Intensified-CCD (ICCD) image sequences were used with in situ laser reflectivity (top) to predict thin film growth kinetics (bottom).
Scientific Achievement
A machine learning model was trained with in situ plasma plume imaging diagnostics to predict the kinetics of thin film growth during pulsed laser deposition (PLD).
Significance and Impact
This work demonstrates that plume imaging can be used to improve the reproducibility of film growth with PLD as well as potentially form the basis of a predictive, model-based control system for guiding autonomous synthesis on-the-fly.
Research Details
- A (2+1)D convolutional neural network (CNN) was used to extract complex spatiotemporal features from plasma plume dynamics to predict film growth kinetic parameters.
- The CNN was trained to predict the flux, J, and sticking coefficients, s0 and s1, for a film growth model that are derived from in situ laser reflectivity monitoring.
S. B. Harris, et al., npj Computational Materials 10, 105 (2024). DOI: 10.1038/s41524-024-01275-w
Work was performed at the Center for Nanophase Materials Sciences
